RMP.3:4.4 - Preserve useful commitments and method-appropriate flexibility
Record what was decided, why, and its timing relative to relevant data access, outcome exposure or interpretive engagement. Distinguish a study registration, a protocol and an analysis plan by their actual content. The label “preregistered” alone does not tell a critic which choice preceded which information.
Commit in advance where the chosen Method and receiving use rely on that commitment. Permit bounded evolution where it belongs to the inquiry, and explain how a change remains connected to the same question. If an external institution requires a particular registration or approval, obtain that result through its own rules; RMP does not grant it.
For example, a reflexive inquiry into how former workers experienced a factory closure can refine a prompt after an encounter exposes a previously unnoticed meaning. Preserve the starting prompt, the reason for refinement and the affected encounters. Changing the task to estimating the fraction of all former workers who agree with a proposition requires a different question and design; calling it another interview amendment does not preserve the original inquiry.
Other design families require their own operational distinctions:
| Inquiry | Consequential detail and return |
|---|---|
| Historical or source-critical | State which inspected documents can establish chronology and how contrary or inaccessible sources affect it. An unavailable earlier series cannot be treated as inspected silence. Return a defeated chronology comparison to RMP.2. |
| Computational or model-based | Specify the model–target relation, input and parameter choices, transformations, environment needs and intended sensitivity or analysis Method. A successful program run cannot supply a missing model–phenomenon justification. |
| Evidence synthesis | Make the supplied review question actionable through eligibility, search arrangements, screening, extraction and appraisal intentions. A convenient source restriction that can change the answer reopens the affected design; reporting the restriction is not sufficient justification. Actual synthesis remains separately performed. |
| Mixed inquiry | Explain what each component contributes, what they share, and how their results will be related. Check shared access, participants, resources and assumptions. Losing an archival chronology component cannot be repaired merely by adding interviews that depend on the same retrospective account. |
These are examples of method-sensitive choices, not a closed taxonomy or a checklist applied in full to each study.